rhce-log-030528.txt

--- Log opened Wed May 28 19:55:46 2003 -!- sweeper [~mbroome@moya.trilug.org] has joined #trilug-rhce -!- Topic for #trilug-rhce: Next meeting - Wed May 21st at 8pm - right here in IRC -!- Topic set by jeremyp [] [Wed May 28 11:42:42 2003] [Users #trilug-rhce] [@ChanServ] [ jimstigator] [ Sinner_] [ Tribot ] [ jeremyp ] [ jtower ] [ sweeper] [ tripaste] -!- Irssi: #trilug-rhce: Total of 8 nicks [1 ops, 0 halfops, 0 voices, 7 normal] -!- Channel #trilug-rhce created Sun Apr 6 17:01:25 2003 -!- Irssi: Join to #trilug-rhce was synced in 0 secs --- Log closed Wed May 28 19:56:01 2003 --- Log opened Wed May 28 19:56:42 2003 -!- sweeper [~mbroome@moya.trilug.org] has joined #trilug-rhce -!- Topic for #trilug-rhce: Next meeting - Wed May 21st at 8pm - right here in IRC -!- Topic set by jeremyp [] [Wed May 28 11:42:42 2003] [Users #trilug-rhce] [@ChanServ] [ jimstigator] [ Sinner_] [ Tribot ] [ jeremyp ] [ jtower ] [ sweeper] [ tripaste] -!- Irssi: #trilug-rhce: Total of 8 nicks [1 ops, 0 halfops, 0 voices, 7 normal] -!- Channel #trilug-rhce created Sun Apr 6 17:01:25 2003 -!- Irssi: Join to #trilug-rhce was synced in 1 secs < Sinner_> sweeper is known as sweeper-boomerang < sweeper> that should be sweeper-broomerang :) < Sinner_> XDDDDD < sweeper> had to restart using screen since I'll prolly be switching to a different machine later < Sinner_> today looks like a good for FOSS < sweeper> tribot: what is FOSS? < Tribot> wish i knew, sweeper < Sinner_> Free and Open Source Software * jtower considers moving to munich < sweeper> yes yes yes. good day < Sinner_> Tribot: learn FOSS is Free and Open Source Software < Sinner_> jtower: beer, Linux ünd funkengruven < jtower> and a bmw < Sinner_> btw, the CEO of microsoft.de has resigned -!- gmontag [~gmontag@cpe-024-211-155-213.nc.rr.com] has joined #trilug-rhce < Sinner_> jtower: and yes, a bmw < jtower> 2 bmws - a two wheeler and a four wheeler < Sinner_> that's my man -!- jtate_ [~chatzilla@rdu74-181-041.nc.rr.com] has joined #trilug-rhce < jtower> i drive past capital bmw almost every day -!- jtate_ is now known as jtate < jtower> some of the new models are very funky looking < Sinner_> ahhhhh, such a view.... < Sinner_> jtower: specialy from behind. The new rear lights are very... bizarre < jtower> i miss my old gpz550 < Sinner_> and the newest 850 looks just like a Ford Sabre or something * jtate thinks that the 3 series look too much like honda civics from behind < Sinner_> honda vicis are being made to look just like 325 < jtower> the 7 series is just an abomination < Sinner_> the new one is < Sinner_> (deleted) < jtower> so ugly it looks like GM designed < jtower> it < jtower> no automaker even comes to close to GM for god-awful designs < Sinner_> do you know if BMW still makes the M5i ? < jtower> what's the i? < Sinner_> a badder & bigger engine < jtower> far as i know there is only one us model -!- jtate_ [~jtate@rdu74-181-041.nc.rr.com] has joined #trilug-rhce -!- RedWolfe [~chatzilla@durham-ar1-4-64-250-023.durham.dsl-verizon.net] has joined #trilug-rhce < jtower> 4.4L V8, 400hp < Sinner_> the "sporsty" 5-series < Sinner_> sounds like it < jtower> i'm actually considering getting a used E36 M3 if i can find one at the right price < RedWolfe> tribot: excuse < Tribot> RedWolfe: excess surge protection < Sinner_> jtower: are you looking for sppeding tickets? >:) < jtower> actually, i've been ticket free for nearly three years < jtower> that's a record for me < Sinner_> how do you do that? < jtower> work at home, and never leave :) < gmontag> I have a computer power question when my dishwasher or washing mashine or other 220v appliance comes on it makes make my screen shimmer < gmontag> how do I get rid of this < Sinner_> gmontag: get a surge protector on the PC / get a little further from the dishwasher / plug the PC toa different wall outlet, one that does not share same circuit with d/w < jtower> google for 'non-shimmy monitor' < gmontag> tried surge protector and even ups. It happens to numerous outlets < jtower> call an excorcist < RedWolfe> often implies faulty ground circuitry in the house < Sinner_> cheaper: play "Tubular Bells" < jtower> Tribot: divine will gmontag fix his monitor problem * Tribot shakes the psychic black sphere... < Tribot> It says 'My Sources Say No,' jtower < jtower> gmontag: you're SOL < jtower> you should quit now while you're ahead < gmontag> google said no results that might be a record < jtower> no, the trick is to get exactly ONE hit -!- pihlopase [~pihlopase@rdu26-245-048.nc.rr.com] has joined #trilug-rhce < jtower> that's a lot more difficult < gmontag> I moved the systems to another room but I wanted to set up a system as a dvr in the other < jtower> anyone here have a bunch of spam stored in a folder? < Sinner_> gmontag: what about wireless power? :P < jtower> power over ethernet < Sinner_> jtower: my /dev/null is about to fill up < jtate_> yeah < jtate_> jtower: I've got a doozy. 900+ messages. < RedWolfe> jtow: i throw it out as soon as I sa-learn it < jtower> jtate_: whenever you get a chance can you send a bunch of spam to test1@sfme.com < jtower> i'm trying to test out a spamassassin install < jtower> no rush < jtower> i see stuff from sinner flooding in :) < jtate_> How are you doing that Sinner_? < Sinner_> jtower: I fwd you a couple, fresh ones < jeremyp> Evening folks. Sorry I'm late. < Sinner_> jtate: not bad. Peter Frampton, lemonade... < Sinner_> how-dee jeremyp < jtower> late, and he doesn't like monty python < jtate_> A lot of the mail I have in there is already tagged with spamassassin. So it's either modified by SA or it's an attachment. < sweeper> jeremyp: howdy < jtower> jtate_: ah, that won't work then * jtate_ wishes jtate would leave so I can have my nick back. < jeremyp> jtate_: I'm pretty sure sa-learn recognizes if it's already tagged < jtate_> Yeah, it does, but he wants fresh spam to test his sa rules. < jeremyp> jtate_: you haven't registered your nick? shame shame :) < jeremyp> Ah. There's a way to forward spam through spamassassin to remove the headers/tags/what have ya < jeremyp> spamassassin -d -!- jtate [~chatzilla@rdu74-181-041.nc.rr.com] has quit [Read error: 110 (Connection timed out)] < jeremyp> it expects things one at a time, so if you have a whole mbox file you'd need to use formail < jtate_> I had a goof using chatzilla. So now that I'm connected using xchat. There he goes. -!- jtate_ is now known as jtate < jeremyp> If you register your nick you can kill that kind of thing right away with /msg nickserv ghost < jtate> Ok, so how does one register their nick? < Sinner_> jeremyp: any register-your-nick-HOWTO ? < jeremyp> type /msg nickserv help < jtower> Sinner_: only one of those scored >5 < jtower> does forwarding lower teh SA score? < jtower> i assume it does to some extent < jeremyp> So, < Sinner_> jtower: maybe FWD lowers de spamindex -!- ccw [~ccw@durham-ar1-4-64-250-023.durham.dsl-verizon.net] has joined #trilug-rhce < sweeper> jtower: I'm pretty sure it does. I'll bounce a few to you. all that does it add a few more received headers < ccw> who am us, anyway ... < jeremyp> You can't forward messages and expect the headers to be anywhere near intact. < jtower> thanks sweeper < jeremyp> But yes the "bounce" feature of pine is useful < jeremyp> Though as Tanner and I learned recently, the bounce feature still adds some headers to indicate who it's "resent" from < jtower> jeremyp: yeah, but header checks are only a part of the total SA score < jeremyp> so if you have Bayes or something going, then it will notice those < Sinner_> Tribot: whois ccw < Tribot> it has been said that ccw is Chris Woodbury < Sinner_> there you go < jeremyp> There are a lot of header checks though. THey are more important than body checks usually. < jeremyp> Which is why the forwarded messages didn't score very high, since only the body checks worked right < ccw> Tribot: that's Christopher C. Woodbury, to you! < jtower> ah. i haven't actually gone through the rules and weighting one by one < Sinner_> i Christopher X. oodbury < jtower> i tried signing up my test account for spam last night, but i haven't realyl gotten much < Sinner_> duh < Sinner_> i Christopher C. Woodbury < jtower> who would have thought it was difficult to start getting spam? < Sinner_> mumble...spell checker...old keyboard...mumble < sweeper> jtower: I bounced you 10 that my ISP's SA install didn't tag < jtower> thanks mike < jtate> Tribot: ccw is Christopher C. Woodbury < Tribot> ...but ccw is Chris Woodbury... < RedWolfe> you want spam? go post on news.admin.net-abuse.email :-) < jtower> sweeper: your messages were >5 on 9/10 < jeremyp> Tribot: no, ccw is Christopher C. Woodbury < RedWolfe> tribot whois ccw < Tribot> ccw is Christopher C. Woodbury < jtate> jeremyp: I told tribot to forget ccw privately < jeremyp> Okay, so who has questions about kernel compiling / kernel services, etc? < jtower> i have no questions about /etc < sweeper> jtower: cool. lemme know if you need moew < jeremyp> jtate: ah. You can use "no, " to make it overwrite a previous factoid; you don't need to forget it first < jtower> sweeper: sure, keep em coming. the more the better < gmontag> If you patch a kernel do you have to compile it and void the support from redhat -!- mode/#trilug-rhce [+o jeremyp] by ChanServ -!- jtower was kicked from #trilug-rhce by jeremyp [grr....] -!- mode/#trilug-rhce [-o jeremyp] by ChanServ -!- jtower [~jason@moya.trilug.org] has joined #trilug-rhce -!- Jaimie [~jaimieliv@rdu163-56-231.nc.rr.com] has joined #trilug-rhce < jeremyp> gmontag: well, most of us have little "support" from Red Hat anyway, but yes, if you compile your own kernel you have no more support < jeremyp> (on kernel related things, yes) < gmontag> so you have to recompile to apply a patch < jtate> gmontag: yes. Now if it's just a module, you can compile it separately. < jtower> dumb question: will the result of building a src.rpm differ if two different kernel versions are used? or does that depend only on the version of gcc used to make them? < ccw> sweeper: please accept offered file ... < jeremyp> gmontag: if it's a Red Hat supported "patch" they will distribute RPMs for it < jeremyp> But many things are modules, not patches. A rare exception is Win4Lin which requires a patched kernel. < sweeper> ccw: trying to figure out how ... < jeremyp> ccw: "/dcc receive" maybe ? < Sinner_> jeremyp: doesWin4Lin kernel break RedHat support? < gmontag> The specifics is there is an acpi patch which will allow my laptop to suspend < jeremyp> Sinner_: if you have a support contract with Red Hat (ie, you use Red Hat Enterprise Linux, then installing a Win4Lin kernel would void your support contract, yes.) < jeremyp> ADPI is most certainly not supported by Red Hat < jeremyp> ACPI < sweeper> ccw: got it (it's /dcc get [nick [file]] < jeremyp> As is pointed out in the Red Hat Linux 9 release notes < Sinner_> jeremyp: thanks for the info < jeremyp> Which I really hope everyone here has read by now! < jeremyp> Release notes are critical documents to read. < Sinner_> ACPI is a piece of... of... cow manure -!- Nat_RH [~Nat@rdu74-184-217.nc.rr.com] has joined #trilug-rhce < jeremyp> gmontag: but you should be able to re-enable ACPI pretty easily if you look in the kernel src.rpm (SRPM). < jeremyp> Because it was there during one of the beta cycles < jeremyp> But don't expect it to be stable.. there are good reasons why it wasn't included in the final release < Jaimie> Hi, All. Has Jon been around tonite? < Sinner_> I know one: it's not all stable < Sinner_> Jaimie: jcarnes? < Jaimie> Yah. < jtate> cybertooth hasn't checked in yet. < jeremyp> He left a while ago saying Daddy Duty was calling. < gmontag> toshiba laptop needs a kernel patch to enable suspend and show battery power < gmontag> i may just wait for 2.6 < jeremyp> gmontag: but if you don't have a support contract with Red Hat that you're trying to protect, go for it < jeremyp> just be sure to leave the old kernel in your bootloader so you can revert back to it WHEN you have problems. :) < jeremyp> notice that's not an IF when it comes to ACPI :) * Sinner_ notices that jeremyp also knows about ACPI firsthand... * jtate wishes you were all here to partake of the cookies made by my wonderful wife < gmontag> I know the acpi is buggy. I have everything working fairly nice so now I am leary of change. < jeremyp> So you are using the ACPI patches then? < Sinner_> I tried to use them. no luck < gmontag> no I enabled apm which conflicts with acpi and installed software written by a jonathan buzzard which allows me to manually control my fan < jeremyp> Ah, cool. < jeremyp> If your BIOS has apm support then by all means use it instead. < gmontag> fan -n and my laptop gets cooler < jtate> so to build your own kernel from the kernel-source rpm: < jtate> copy the proper config file to .config < jtate> make oldconfig < jtate> make dep < jtate> make clean < jtate> make bzImage < jtate> make modules < jtate> make modules_install < jtate> and finally make install? < jeremyp> you left off make mrproper, which you do first after installing the kernel-source rpm < jeremyp> but otherwise looks good < jtate> Right. I'm confused about the make clean after make dep < jtate> How do you completely clean your build tree? < jeremyp> Also, you make want to edit the Makefile to change the release tag < RedWolfe> make mrproper is the "distclean" for the kernel < jtate> Yeah, but that's optional. < jeremyp> "make mrproper" completely cleans out everything < gmontag> is there also something to back up your modules < jeremyp> including your .config < jeremyp> gmontag: just make sure you're not using the exact same name-version-release as the existing kernel and you don't need to backup anything < jeremyp> the default for the kernel-source RPM will append a "custom" to the release tag < jeremyp> I usually change that to 'jpp' or something so I remmeber it's my kernel config < jeremyp> jtate: about the make mrproper; it's necessary because sometimes the kernel-source RPM doesn't "ship" with a cleaned out build tree in it. I dunno why. < jtate> Cause someone forgot to run it before they tarred it up probably. < jeremyp> Yeah, but apparently there's some more complex reason related to the Red Hat build system < jtate> Although you think they might put that in the SPEC file. < jeremyp> It's because the same tarball is used to generate the kernel-source RPM that generates the binary kernels, I think' < jtate> Yeah, the src.rpm file generates the kernel-source rpm < Sinner_> and someone generates the spec file -!- RedWolfe changed the topic of #trilug-rhce to: Next meeting - Wed May 28 2003 at 8pm - here, same Bat Time, same Bat Channel < Sinner_> has anyone tried to compile the kernel with the ncurses utility yet? < jeremyp> make menuconfig? < jeremyp> I used to use it years ago. But xconfig is generally nicer. < jtate> I'm doing that right now. < Sinner_> jeremyp: what about when there's no X? Like, say, in a live test? < jeremyp> Well sure menuconfig (or just "make config") works fine. < jeremyp> What do you mean by "live test" though? * Sinner_ thinks that is good to be familiar with it < Sinner_> live test = RHCE exam < RedWolfe> for a while in 7.x the menuconfig was borken * jtate doubts they're gong to make us compile a kernel in the test * Sinner_ believes that it will be in the test -!- cybertooth [~cybertoot@rdu57-28-206.nc.rr.com] has joined #trilug-rhce < cybertooth> Did I mis anything exciting? < jeremyp> Yes. < Sinner_> cybertooth: we were ahcking into SCO servers, freeing the information < cybertooth> darn... and I had a hankering to talk about apc tonight < jeremyp> apc? < jtate> Go ahead. Don't think anyone but jeremyp could stop you < jeremyp> American Power Conversion, the UPS maker? They're cool. < cybertooth> advanced power control < jeremyp> apm ? < jeremyp> We were talking about apm and acpi earlier if that's what you mean. < jeremyp> Mostly that ACPI's linux support sucks < cybertooth> X10 < cybertooth> What do you mean it sucks? RH9 does great on my Satellite Pro < jeremyp> Not with ACPI. RHL 9 doesn't do ACPI < jeremyp> You're using apm on that older laptop, which is fine. < jeremyp> Too many acronyms, eh? :) < cybertooth> Yep, it works fine. < cybertooth> So the newer bioses dont support apm they only support acpi < jeremyp> Some of them, yes. Some BIOSes support both < cybertooth> Now I wish I were hacking into sco servers and freeing the information... < cybertooth> ... you know most phone switches are still using sco < Sinner_> see? < cybertooth> ACD's as well. < jeremyp> cybertooth: grr... :) < Sinner_> maybe this is why Sprint is moving to IP networks... and other telcos are swithcing to LinuxCarrierGrade * RedWolfe sings "info everywhere just gotta be free....." < cybertooth> IP networks make a lot more sense for these folks. The added portability of moving to IP is fantastic, plus it simplifies the network. < cybertooth> Also failover becomes much easier. I love telephony < jtate> drm_os_linux.h:16:2: warning: #warning the author of this code needs to read up on list_entry < cybertooth> I see a day in the very near future where the phone plugs into a network socket and simply works. * Sinner_ wonders how dooes one reboot a phone after installing a new kernel with caller-ID support < cybertooth> How about simply running it off a USB mem chip? < RedWolfe> you'd just download a new module, not reboot. < jeremyp> jtate: heh < cybertooth> If you latency to the switch was low, you could move the phone to *any* network connection and simply plug in to make/recieve calls < cybertooth> Like a mobile phone now, but much cheaper and more reliable. < RedWolfe> to reboot, just unplug and replug, it would "bootp" itself (like the "merlin" did < cybertooth> Tech support leaves for the night and takes the phone home with them to plug into their network connect at home - since it's their turn to do night duty. < Sinner_> what's best phone os loader ? lilo? grub? < Jaimie> If telco nets go to IP, what happens to dial-up services. Would the 53Kbps limit imposed by the FCC go away? < cybertooth> oddly enough, it's the protocol used that puts that limitation on the line. < cybertooth> The same line with a different protocol can do up to 15Mb. < sweeper> it's the robbed-bit signalling < jtower> remember, sprint moving to IP networks is not the same as voice over the internet < cybertooth> Also the CLEC would need to be modified to handle the protocol difference. < cybertooth> As long as folks will pay for 56kb dial-up, there's no big motivation for the telco's to modernize the old switches. < RedWolfe> the voice channel is a tiny part of the copper pair bandwidth < cybertooth> With QoS installed, and if you stay inside the infrastructure of a large carrier, then you should be able to do VoIP all across America. < Sinner_> large carrier? CVN-Carl Vinson? < cybertooth> As long as the latency is less than about 25ms then the call sounds okay. When the latency crawls up to over 50ms then it's like talking over a bad speaker phone. < cybertooth> Large carrier: I'm thinking TWTC < cybertooth> ... or RoadRunner. < cybertooth> You could have a gazillian offices all over america all controlled by one centrilized phone switch (or switches for rundunancy) < jeremyp> So, any more kernel comments / questions? < jeremyp> Anyone have any comments on Software RAID? < cybertooth> Then your folks could simply plug into any RR connection in America and get their calls. < cybertooth> jeremyp: subtle :-) < jtate> jeremyp: still building mine < RedWolfe> have to use IPv6 for enough identities < jtower> dumb question: will the result of building a src.rpm differ if two different kernel versions are used? or < jtower> does that depend only on the version of gcc used to make them? < jtower> to repost my earlier question :) < jeremyp> jtate: yeah, it can take a while if you don't go through and turn of the gadzillion modules (like I was doing last wed. night) < Sinner_> src.rpm != skernel-source.rpm < jeremyp> s/of/off < RedWolfe> well, of course two different versions are going to contain different code... < jtower> i mean a regular old src.rpm, not a kernel rpm or anything < jeremyp> jtate: not sure exactly what you mean. There's some special mojo to make the src.rpm build the different types of kernels (e.g. i386 vs. athlon, bigmem, etc) < jeremyp> sorry jtate , meant jtower < Sinner_> jtower: they should have no differences. < jtower> i'm not talking about building kernels, just regular packages < jtate> jtower: I don't think so. Now if you have different versions of glibc then that's a different story. < RedWolfe> oh, ok, the resulting binary *may* be usable by different kernel versions, it depends on several factors < jtower> had a case a couple of weeks ago where i ran into this < jeremyp> jtower: ah, yeah, what Sinner_ said. It's all about glibc and the other factors < Sinner_> jtower: I say that, if glibc is the same, kernel versions do not matter < sweeper> jtower: yeah, there's the mojo for the architecture, and if the source includes the kernel headers, it could end up with a different result < jeremyp> When you're building kernel *modules* it matters what kernel source version is installed. But it doesn't matter which one is actually *running* < jtower> it was a module < Sinner_> jtower: ... except if you change kernel versions (2.4/2.4/2.5...). < jeremyp> See a kernel module is not "just regular packages" < jtower> jeremyp: good point < RedWolfe> the main reason for RH8 vs RH9 was that the binary compatibility changed < jtower> i meant regular package != kernel package < jeremyp> Depending on how the kernel module's build environment is set up, you probably have the option to point it at different directories for the kernel source. < RedWolfe> a *kernel* module would probably count as a "kernel package" < jtower> it was a weird case, HP supplied a src rpm for a gigabit NIC < jeremyp> So if you have multiple kernel-source packages installed you could pint it there. < jtate> That's different because it's a kernel module. < jtower> then the kernel was updated by up2date (RHEL AS) < jtate> Then you've got to rebuild the source package. < jeremyp> So you'd need to recompile with that new kernel < jtower> and things got all messed up. i didn't handle it, someone else did. i just helped fix the problem < jtate> Hey 12 minutes. < jtate> to make modules < Sinner_> ram & cpu ? < RedWolfe> jtate: nice fast machine there < jeremyp> Right, not recompile the kernel itself, just recompile that module to match the new kernel source < jtate> 512MB on an XP 2100+ < jtower> the problem was, it was a remote server. messed up NIC driver/module makes it hard to ssh in < jeremyp> Uh, yeah. THat's why you don't upgrade kernels remotely/automatically. < jtate> Yeah that would make it difficult. < jeremyp> up2date won't do it by default. Someone must have messed with the up2date config. < jtower> i was trying to "reverse engineer" what they did wrong < jeremyp> it excludes kernels < jtower> jeremyp: that's exactly what happened < jeremyp> Whoever did that needs a kick to their rump. < RedWolfe> (usually excludes kernels) < jtower> plus on top of that, another HP package modified the bootloader < jeremyp> Especially since they were using a non-standard module that needed to be compiled < jeremyp> (if you have all standard, well-supported drivers, it could be less of a problem.) < jtower> believe me, no more HP servers for me, thanks very much < jeremyp> Yeah even if you do install kernels with up2date, I think it leaves the old one as default in the bootloader. < jeremyp> I'd have to check on that to be sure. I'm pretty sure it does that with an 'rpm -i' install so up2date should do the same, since the underlying RPMs are the same. < jtate> Yum doesn't. It puts the new kernel as the default. < RedWolfe> no, if you do up2date the kernel, the GRUB/LILO is modified to use the new kernel < jtower> i think what happened is they installed the new kernel but didn't reboot. then they rebuilt the nic module using the wrong tree < jtate> So I think that you're mistaken. < jeremyp> Hm. I could see why you'd want the new kernel as default in some environments. < RedWolfe> (It's got the new security patches,, gotta have those! :) < jtate> I've heard talk of an update that was pulled recently by RH. Which package was it? < jeremyp> A kernel package that broke a certain sound card driver. < jeremyp> Not a big deal. < jtate> Which version? < jtate> kernel and distro? < jtower> jeremyp: unless you're blind :) < RedWolfe> didn't hear of a RH fumble, M$ fumbled an XP patch < jeremyp> I'd have to check to see the specifics... hold on. I just remember seeing something about it on shrike-list < jtate> Anybody know if the Soundblaster Platinum Live 5.1 is supported? < jtower> the SB live cards are well supported < jtower> including the 5.1 < RedWolfe> tribot: insult < jtate> It doesn't work on my RH9 system. < jtower> you may not get all of the fancy-scmancy features that the windows drivers provide though < jtower> what motherboard? < jtate> It's an Asus KT333 < RedWolfe> tribot: insult ccw < Tribot> problem connecting to "insulthost.colorado.edu", port 1695: Connection refused < jtower> there is a known conflict between the via 686b southbridge and SB live cards < jtower> or so i've read < jtate> It works reasonably well in Winbloze < jtower> try a different pci slot, sometimes pci busmastering gets in the way < jeremyp> http://www.redhat.com/archives/shrike-list/2003-May/msg01152.html < jeremyp> http://www.redhat.com/archives/shrike-list/2003-May/msg01375.html < RedWolfe> jeremyp: summary please? < jeremyp> http://www.redhat.com/archives/shrike-list/2003-May/thread.html <-- search for "kernel-2.4.20-13.9 Breaks Soundcard, Mozilla, ETH0" < jeremyp> I don't know for sure if that update (2.4.20-13.9) has been "pulled" but they acknowledged there were mistakes made in the SB Audigy driver < jeremyp> that second URL I posted has some test kernels that fix it * jtate has a platinum not an audigy * ccw contimplates the appropriate insult for RedWolfe... < jtower> jtate: also, try changing the PnP setting in the bios < jtower> sometimes that clears up irq problems < ccw> btw, ccw has tribot ignored ... < jeremyp> In one of those threads somone from Red Hat acknowledged they didn't have an Audigy in their test lab. < jeremyp> They said that had been rectified after this big mistake! < jtate> jtower: will do. < jeremyp> ccw: don't worry, the insult server is broken anyway < jtate> Have to wait for the updates to install. < ccw> isn't that nice... < jtate> jeremyp: Ok, I've done a make install, and I have the vmlinux in /boot, but there's no initrd. Is there a make step to do that, or is that a manual process? -!- scottl [~scott@user-0c8hpp4.cable.mindspring.com] has joined #trilug-rhce -!- lovelace [~lovelace@moya.trilug.org] has joined #trilug-rhce < jeremyp> jtate: depends, if you have no need for one, it might not have created one < jtate> Hmm. < jeremyp> Did you compile ext3 into the kernel? < jtate> What's the System.map file? < jtate> Yes < jeremyp> Ah, then you probably don't need an initrd. < jeremyp> Usually the ext3 stuff is in the initrd < jeremyp> because it's compiled as a module to keep the kernel size down < jtate> I also compiled LVM and RAID1/0 in. < jeremyp> Hmm, LVM without initrd might eb interesting. Ask jtower about that. < jtower> i built a server a couple of days ago w/ software raid 1 and LVM < jeremyp> But the answer to how to make it manually is the "mkinitrd" command (see man page) < jtower> and the installer said it couldn't make a boot disk < jtate> Yeah, I know that. I got an error doing that. < RedWolfe> mkinitrd is a separate step < jeremyp> RedWolfe: no, it's part of "make install" in the Red Hat provided kernel source tree < jtower> not enough space for the kernel and initrd i presume < jeremyp> right < jtower> just an FYI < jeremyp> unless you use a 2.88 MB boot disk, but convincing mkbootdisk to do that might be interesting < jtower> software raid or lvm alone will fit, but not both < Sinner_> ok guys < Sinner_> good night < Sinner_> (someone save the log) < jtate> I'll just use tom's root boot if I need to. < RedWolfe> "make install"? when did they add that, RH8? -!- Sinner_ [~sinner@ip68-101-8-79.nc.hr.cox.net] has quit ["gotta catch some sleep"] < jeremyp> jtate: the problem with LVM compiled into the kernel (instead of as a module/initrd) is there are a few user-space commands that need to be run to init. the LVM array < jtate> Doh. < jeremyp> So you must use modules and the initrd if you want / to be mounted on LVM < jtate> Cannot open root device 3a02 < jtate> kernel panic < jeremyp> (and /boot still can't be on the LVM, because it's got to read the initrd from somewhere) < jeremyp> Yeah, looks like that's the problem. You do need an initrd. < jtate> No, /boot isn't there. < jeremyp> But / is, correct ? < jtate> It's raid0 < jtate> Yes < cybertooth> Technically speaking you *could* use a cd to boot it all < scottl> ergo dangerous to have / and /boot on LVM correct? < cybertooth> just /boot < jeremyp> cybertooth: yeah, if you're an isolinux guru < cybertooth> just catching up on the scroll < jeremyp> scottl: we were talking about the setup necessary for / to be on LVM. /boot cannot be on LVM at all. < RedWolfe> but /boot *can* (apparently) be on a raid0 device < jeremyp> RedWolfe: make install has been in the kernel Makefile forever. I don't know when Red Hat put the mkinitrd step in it, but probably before RHL 8.0 * jtate swears at his router again < jeremyp> RedWolfe: ITYM RAID 1 < RedWolfe> mirrored < jeremyp> RAID 0 is striping with no parity (no fault tolerance) and is typically used with video data or other high-speed stuff < jeremyp> RAID 1 is mirroring. < RedWolfe> ok, raid1 -!- scottl is now known as scott < jeremyp> And zero does not equal one. *grin* < RedWolfe> Thanks, folks, TTFN < jeremyp> RedWolfe: good night -!- RedWolfe [~chatzilla@durham-ar1-4-64-250-023.durham.dsl-verizon.net] has left #trilug-rhce [] < jeremyp> jtate: you understand what we were saying about the initrd being necessary for LVM, right? If you recall how I "took apart" the initrd at the meeting, you can do that again and see what it does exactly. < jeremyp> (gunzip the .img file, and then mount it lookback) < jeremyp> loopback < jtate> I don't understand why it needs to be in the initrd if it's compiled into the kernel. < jtate> Oh, because some LVM commands have to be run. < jtate> by init < cybertooth> ...before it can get to the kernel < cybertooth> ...which is stored on lvm < jeremyp> Well, before it can get to the / filesystem is the important part. < jeremyp> The kernel itself is already loaded. < jeremyp> But the LVM user-space commands are normally on the / filesystem. < jtate> But the kernel is in /boot < jtate> Right. Gotcha < jtate> So, the module has to be in initrd so that init can read the / filesystem to run init. < jtate> So, the module has to be in initrd so that init can read the / filesystem to run those commands. < jtate> Before / is mounted < jeremyp> No. < jeremyp> The commands themselves are in the initrd image, which remember is a ramdisk. < jtate> Duh. Sorry. I'm feeling pretty dense tonight. < jeremyp> The ramdisk is loaded (copied) by the kernel from /boot/initrd-<blah>.img < jtower> you see, chickens come from eggs... < jtate> Ok, so if I've got lvm compiled into the kernel, how do I get an initrd with just those commands? < jeremyp> Because the / filesystem cannot be read if the vginit (or whatever it is) isn't run. < jeremyp> jtate: now that's a good question! :) < jtate> mkinitrd bombs out on me because it's looking for the raid0 module. < jeremyp> There's a --no-raid option or something. Check the man page. < jeremyp> remember that mkinitrd is looking at your /etc/modules.conf to determine what to build < jtate> It's not in /etc/modules either < jeremyp> so you can run into problems if those modules don't exist. (which they don't, because you just built those things in the kernel instead of as modules) < cybertooth> Good Luck - < jeremyp> It might do some other mojo if it sees you're running software RAID. See the man page though. -!- cybertooth [~cybertoot@rdu57-28-206.nc.rr.com] has left #trilug-rhce ["Client exiting"] < jeremyp> though the man page unfortunately doesn't mention LVM :/ < jtate> --omit-raid-modules --omit-lvm-modules < jeremyp> Doh, I was looking at the man page on a 7.3 box! < jeremyp> Which doesn't have LVM support. < jtate> Hehe. I'll reboot and try that. < jeremyp> Ah, I see... mkinitrd looks at fstab and raidtab as well as modules.conf < jtate> Looks good so far < jeremyp> There's also a --builtin option to mkinitrd so you can specify that a given parameter is built in < jeremyp> (that probably mostly applies to device drivers, not so much the RAID or LVM support) < jtate> It just booted without a hitch. < jeremyp> Great. < jtate> Now to get the sound stuff working < jeremyp> Heh can't help you there. < jeremyp> I still can't get sound working on zero out of the three computers I've tried extensively on. < jtate> Yeah, I know. If I have problems, I'll take it to the trilug. < jeremyp> I've decided I'm just doomed to no sound in Linux. I've learned to live without it. -!- jtate [~jtate@rdu74-181-041.nc.rr.com] has left #trilug-rhce ["Noone's here anymore"] < jeremyp> Heh. The rest of us are apparently relegated to "noone" status :) < jeremyp> lovelace: gonna get a word in edgewise or just lurk? :) < Jaimie> quick, quick - what's the command line to search for and replace a text string in a series of files in multiple recursed subdirectories. * lovelace lurks < jeremyp> Jaimie: I've seen someone post perl mojo that does stuff like that before. < jeremyp> Jaimie: ask on #trilug < Jaimie> I was just seeing who was still awake. < jeremyp> Heh. Not many apparently. < Jaimie> Guess not. I'm to be. Night-Night -!- Jaimie [~jaimieliv@rdu163-56-231.nc.rr.com] has quit [] --- Log closed Wed May 28 23:14:42 2003 Generated by irclog2html.pl 2.1 by Jeff Waugh - find it at freshmeat.net!